The experiments proposed herein are designed to explore the efficacy and mechanics of progressive relaxation training and benzodiazepine therapy in improving glucose metabolism in both insulin dependent (IDDM) and non-insulin dependent diabetes mellitus (NIDDM). Twenty-four insulin dependent and twenty-four NIDDM patients will participate in two experiments sequentially. In the first experiment patients will receive a glucose tolerance test (NIDDM only), an insulin tolerance test (NIDDM only), an epinephrine sensitivity test and psychometric assessment. Serial blood samples will also be drawn for glucose determination. Subjects will then be randomly assigned to receive alprazolam, a benzodiazepine, or matched placebo. After 8 weeks, the glucose tolerance test and insulin tolerance test for patients with non-insulin dependent diabetes will be repeated. Blood samples will also be drawn for glycohemoglobin measurement. The drug condition of patients will then be reversed and subjects will be reevaluated 8 weeks later. Four weeks following the termination of the drug study, patients will be readmitted to the hospital during which time the glucose tolerance test will be performed on NIDDM patients. Serial blood samples for glucose determinations will also be drawn. All patients will then be randomly assigned to either control or treatment conditions: l) progressive relaxation with compliance enhancement instructions and 2) no behavioral treatment. Patients will return every 8 weeks for assessment of glycohemoglobin, fasting and 2-hr postprandial blood glucose. Glucose tolerance tests will be repeated at 24 weeks and 48 weeks. The effects of the pharmacologic and behavioral interventions will be assessed by their affects on incremental glucose area (NIDDM), 3) mean daily blood glucoses, 4) glucose-stimulated insulin secretion (NIDDM), 5) 24-hr urine glucosed, 6) fasting and 2-hr postprandial glucoses, 7) glycohemoglobin and 8) medication requirements (insulin or sulfonylurea). The mechanism of action of these interventions will be assessed by measurement of plasma cortisol, urinary cortisol and urinary catecholamines during hospital admissions and outpatient visits. An attempt will be made to predict responses to both benzodiazepine and relaxation therapies by subjects' responses to the epinephrine sensitivity test and "paper and pencil" psychometric assessment.